Methodology for Determining Compaction Temperatures for Modified Asphalt Binders

D. Experimental Design

1. Aggregates

Three aggregates were used. Two aggregates were variations of an aggregate blend used extensively in a previous Federal Highway Administration (FHWA) study and in two National Cooperative Highway Research Program (NCHRP) studies.^(1-3) The properties of the aggregates are given in table 1. The blending percentages used in the three referenced studies were 61-percent No. 68 diabase, 30-percent No. 10 diabase, 8-percent natural sand, and 1-percent hydrated lime. This blend met the requirements for a Virginia Department of Transportation (VDOT) SM-3 surface mixture gradation.⁽⁴⁾

The current equiviscous principle, which is based on asphalt binder viscosity, gives a single ranking for all asphalt binders. This ranking is independent of other variables, such as aggregate gradation and asphalt binder content. These variables affect the workability of a mixture, but it is assumed that the ranking given by the equiviscous principle remains valid for all mixtures. It is possible that some aggregate characteristic might interact with a polymer-modified asphalt binder so that the ranking is not always constant. If this occurs, then the equiviscous principle based on asphalt binder viscosity cannot be used. Perhaps, a principle based on the viscosities of mastics should be used. What aggregate property, if any, causes an interaction is not known.

Table 1. Aggregate properties for mixtures containing diabase.

NA = not applicable

Based on the results from asphalt binder and mastic tests performed during a previous FHWA study, varying the composition of the minus 75-m aggregate may lead to an interaction.⁽¹⁾ Dynamic shear rheometer tests performed on five asphalt binders gave the following ranking from most to least susceptible to rutting at 58°C: AC-5, AC-10, AC-20, Novophalt, and Styrelf. (The last binder is also called Stylink.) These asphalt binders had Superpave performance grades (PGs) of 58-34, 58-28, 64-22, 76-22, and 82-22, respectively. Pavement rutting tests provided the following ranking: AC-5, AC-10, AC-20, Styrelf, and Novophalt. Mastics consisting of the asphalt binders with the minus 75-m fractions of the aggregates shown in table 1 were also tested by the dynamic shear rheometer at 58°C. Tests using the minus 75-m fraction without hydrated lime were inconclusive. The first set of tests gave the same ranking as the asphalt binders. However, when the tests were repeated, Novophalt and Styrelf switched places, and the ranking agreed with the pavement rutting tests: AC-5, AC-10, AC-20, Styrelf, and Novophalt. Tests on mastics that included the hydrated lime clearly gave the correct ranking, even though the dust-to-binder ratio by volume was constant for all mastics.

Based on these findings, it was decided to vary the composition of the mastic. Note, though, that the above interaction was found at 58°C, which is well below the temperature needed for compaction. The first aggregate type contained no hydrated lime, while the second aggregate type contained 1.25-percent hydrated lime. Details on how the 1.25-percent level was chosen are given in section F of this report. In this report, these two aggregate types, or blends, are referred to as the diabase aggregate with and without hydrated lime.

The third aggregate type consisted of 100-percent limestone aggregate. This aggregate was designated "RC" by the Strategic Highway Research Program. The properties of the aggregate are given in table 2.

2. Asphalt Binders

The initial polymer-modified asphalt binders used in this study were the Novophalt (PG 76-22) and Styrelf (PG 82-22) asphalt binders used in the previous FHWA study.⁽¹⁾ The unmodified PG 58-28 and PG 64-22 asphalt binders were chosen to be the control binders. The Superpave properties of these binders are given in table 3. Additional properties are given in table 4. Other polymer-modified asphalt binders were to be included in the study after the tests on these asphalt binders were completed.

3. Number of Mixing and Compaction Temperatures

It was expected that a minimum of four compaction temperatures would be needed to obtain a range in temperature that does not affect the volumetric properties of a mixture at N-design. The experimental design using four temperatures and three asphalt binders is given in table 5. A full factorial using two replicates required 24 specimens per asphalt binder.

The mixing temperature was held constant for a given asphalt binder. It was not varied with the compaction temperature. The mixing temperature is normally around 10°C higher than the compaction temperature. Thus, the mixing temperatures used in this study may be too high or too low compared to standard practice. This was done so that for a given asphalt binder, changes in the volume of absorbed asphalt binder would be a function of the compaction temperature alone. This eliminated the potentially confounding effect that a variable mixing temperature might have on the volume of absorbed asphalt binder. Even though the mixing temperature was fixed, it was hypothesized that changes in the mixing temperature would have no effect on asphalt binder absorption because the mixing time of 90 s was relatively short compared to the 2 h of short-term oven-aging (STOA), which was done at the compaction temperature. Also, the water absorption for the two diabase blends was low, being 0.8 percent by mixture mass. The water absorption for the limestone aggregate was slightly higher at 1.45 percent. Mixtures were compacted using a realistic mixing temperature when the test data indicated that this was necessary.

E. Equiviscous Temperatures

The kinematic viscosities of an asphalt binder and the equiviscous principle are used to determine mixing and compaction temperatures. The equiviscous mixing temperature is the temperature where the kinematic viscosity of the unaged asphalt binder is 170 ±20 mm²/s.⁽⁵⁾ The equiviscous compaction temperature is the temperature where the kinematic viscosity of the unaged asphalt binder is 280 ±30 mm²/s.⁽⁵⁾ Prior to Superpave, the equiviscous temperatures were determined using capillary viscometers. The kinematic viscosity of an asphalt binder was measured at 135°C, and the absolute viscosity was measured at 60°C. The absolute viscosity was converted to kinematic viscosity. Capillary viscosity tests were performed in this study at 60 and 135°C in accordance with American Association of State Highway and Transportation Officials (AASHTO) test methods T 201 and T 202.⁽⁶⁾ The viscosities determined by these tests are given in table 4. Superpave uses a Brookfield viscometer, AASHTO test method TP4, and American Society for Testing and Materials (ASTM) test method D 4402 to measure the viscosities of asphalt binders.^(5,7) A minimum of five temperatures were used in this study to determine the relationship between viscosity and temperature. The data are given in table 6.

Mixing and compaction temperatures for the asphalt binders are shown in table 7. Two methods were used to calculate the temperatures from the capillary viscosities: (1) log-log kinematic viscosity vs. temperature, and (2) log-log kinematic viscosity vs. log temperature in degrees Kelvin. The first method is generally used by the hot-mix industry, while the second method is the standardized method. Table 7 shows that the Brookfield viscometer gave the highest mixing and compaction temperatures for Novophalt and Styrelf, followed by the capillary tests where "log temperature" was used. All of the temperatures for Novophalt and Styrelf are high. Mixing temperatures for these two binders rarely exceed 180°C, because higher temperatures may lead to blue smoke. VDOT specified a maximum hot-mix plant temperature of 177°C.⁽⁴⁾

Table 2. Aggregate properties for mixtures containing limestone.

Table 3. Superpave asphalt binder properties.

RTFO = Rolling Thin-Film Oven
PAV = Pressure Aging Vessel

Table 4. Other properties of the asphalt binders.

Table 5. Experimental design.

T_X = compaction temperature

Table 6. Brookfield viscosity (mm²/s) vs. test temperature.

Table 7. Equiviscous mixing and compaction temperatures (°C).

Table 7 includes the temperatures used in the previous FHWA study and by VDOT under the heading "Current Practice."⁽¹⁾ These temperatures did not produce smoke in the laboratory or at the hot-mix plant. It was decided to use these temperatures as the starting temperatures. It was thought that temperatures 20°C above and below the starting temperature would lead to different air-void levels. This provided the first three compaction temperatures to be evaluated.

The temperatures determined by the equiviscous principle can be excessively high for polymer-modified asphalt binders. Therefore, mastics were also tested to determine if they could provide usable temperatures. The optimum mixing temperatures for the mastics were determined first. Table 7 shows that the optimum mixing temperatures for the Novophalt and Styrelf mastics were 180 and 175°C. The methodology used to determine these temperatures is documented elsewhere.⁽⁸⁾ The mixing temperature ranges for the mastics containing the two unmodified asphalt binders were relatively wide. Therefore, the mixing temperatures for the unfilled asphalt binders under "Current Practice" were used.

The compaction temperatures for the mastics were determined using the PG 64-22 mastic as the control. The compaction temperature for this mastic was assigned to be 142°C, which was the compaction temperature for the PG 64-22 asphalt binder under "Current Practice." The kinematic viscosity of this mastic was 1500 mm²/s at 142°C. The compaction temperatures for the other mastics were the temperatures that also provided a kinematic viscosity of 1500 mm²/s. These temperatures are included in table 7. The temperatures provided by the Novophalt and Styrelf mastics, like the Novophalt and Styrelf unfilled binders, were high compared to current practice.

Originally, the PG 64-22 asphalt binder and mastic were to be the primary control materials. However, after the mastic tests using this binder were completed, the supply of this asphalt binder dwindled because it had to be used in two important NCHRP studies.^(2-3) The PG 58-28 asphalt binder and mastic were then designated to be the control materials. Based on the mastic test data, this change would have no significant effect on the mastic compaction temperatures given in table 7.

F. Asphalt Mixture Design

1. Selection of N-Design

There were three options for choosing an N-design for the mixtures. The first option was to use an N-design of 100 or 125 gyratory revolutions, which are used to design mixtures for relatively high traffic levels. One hundred revolutions are used for 10 to 30 million equivalent single-axle loads (ESALs) and 125 revolutions are used for more than 30 million ESALs. The advantage of using one of these N-designs is that most polymer-modified asphalt binders will probably be used in pavements subjected to more than 10 million ESALs.

The second option was to use an N-design of 75 revolutions, which is used for medium traffic levels of 0.3 to 3 million ESALs. The potential advantage of using 75 revolutions is that the higher asphalt binder content that generally occurs with the use of 75 revolutions, compared to 100 and 125 revolutions, may give a narrower compaction temperature range. Density may be more sensitive to changes in compaction temperature as the optimum binder content increases. If so, the temperature range for 75 revolutions should be applicable to 100 and 125 revolutions. An N-design of 50 revolutions was eliminated because 50 revolutions is only used to design asphalt mixtures for low-volume roads. A third option was to use two levels of N-design.

The second option using 75 revolutions was chosen for the two mixtures with diabase aggregate. The applicability of the measured temperature range to N-designs of 100 and 125 revolutions was to be investigated using a partial factorial experimental design if necessary. For the mixture with limestone aggregate, 100 revolutions was chosen because the asphalt binder drained from the mixture during STOA using the optimum asphalt binder content for 75 revolutions. The limestone mixture still appeared to be rich in asphalt binder at 100 revolutions, but there was little to no draindown.

2. Optimum Asphalt Binder Content

The optimum asphalt binder content was chosen for each mixture using: (a) the AC-10 (PG 58-28) control asphalt binder; (b) mixing and compaction temperatures of 145 and 137°C, respectively; (c) 2 h of STOA at the compaction temperature of 137°C; (d) N-designs of 75 or 100 revolutions; (e) a 4-percent air-void level; and (f) two replicate specimens per mixture. The asphalt binder content by volume varied from aggregate to aggregate, but it was held constant for a given aggregate.

Table 8 shows selected properties of the mixtures with the diabase aggregate and various percentages of hydrated lime. The dust-to-binder ratio was allowed to vary with the hydrated lime content so that the mastics would be significantly different in terms of composition. The percentage of hydrated lime to be used in the second mixture was 1.25 percent. This was the highest amount of hydrated lime that could be used based on the Superpave specifications. Hydrated lime contents above 1.25 percent would lead to mixtures with voids in the mineral aggregates (VMAs) below the specified minimum allowable level of 13.0 percent, and a dust-to-binder ratio greater than 1.6. Table 8 shows that the optimum asphalt binder contents for the mixtures with 0- and 1.25-percent hydrated lime were 4.6 and 4.1 percent by mass of the mixture.

Note that the VMA decreased by 1.6 percent with the addition of 2.0-percent hydrated lime by total aggregate mass (14.0 vs. 12.4). This quantity of hydrated lime would decrease the volume of asphalt binder by approximately 1.4 percent if it is assumed that the addition of the hydrated lime does not change the total volume of the compacted asphalt mixture. The VMA would drop from 14.0 to 12.6 percent. Therefore, a large portion of the decrease in the VMA with increasing hydrated lime content could be due to the substitution of hydrated lime for asphalt binder.

Table 9 shows selected properties of the mixture with the limestone aggregate. All properties met the Superpave specification. The effective asphalt binder contents of the mixtures with the limestone aggregate and the diabase aggregate with no hydrated lime were close at 10.2 and 10.5 percent by total volume, respectively, although the mixture with limestone appeared to be much richer in asphalt binder content.

G. Analysis of the Data

The response variable used to determine the compaction temperature range was the air voids at N-design. The temperature that provided the design air-void level of 4.0 percent with an allowable error of ±0.5 percent was found for each asphalt binder. Therefore, the acceptable air-void range was 3.5 to 4.5 percent. This range means that up to approximately 0.2-percent asphalt binder by mixture mass would have to be added or removed to obtain a 4.0-percent air-void level. Thus, the total allowable deviation in asphalt binder content was ±0.2 percent by mixture mass. Although the current equiviscous principle should be valid for the PG 58-28 unmodified asphalt binder, various compaction temperatures were used with this binder so that its allowable temperature range could be compared to the ranges for the other asphalt binders.

For each aggregate, the asphalt binders and mastics were ranked according to their average compaction temperature. Each aggregate should provide the same ranking; otherwise, some type of interaction between the materials has occurred. If an interaction is found, then the compaction temperature cannot be determined using the asphalt binder or mastic property that was measured.

The experimental design assumes that all asphalt binders will provide the same effective (non-absorbed) asphalt binder content for a given aggregate at a 4.0-percent air-void level. If the effective asphalt binder contents are not the same, differences in the compaction temperature ranges for the asphalt binders for a given aggregate may partially be related to the differences in the effective asphalt binder contents. If this is found to be true, the asphalt binder contents may have to be adjusted on a trial-and-error basis so that the effective asphalt binder content is constant.

H. Potential Limitations of This Study

• It was assumed in this study that the optimum asphalt binder content for a given aggregate type and gradation should not vary with the type or grade of asphalt binder. However, a methodology for determining the compaction temperature is still needed if it is found through pavement performance studies that the optimum asphalt binder content should vary with the type of asphalt binder.
  

Table 8. Mixture properties at a 4.0-percent air-void level for the diabase aggregate.

Table 9. Mixture properties at a 4.0-percent air-void level for the limestone aggregate.

• Workability during laboratory compaction may not be the same as workability during field compaction for all mixtures. The required compaction temperature may vary with the type of compaction. This study only addresses compaction using the Superpave gyratory compactor, even though laboratory workability vs. field workability is an important issue.
  
  
• This study did not address possible differences in the mechanical properties of laboratory-produced specimens and specimens cut from pavements when the two types of specimens have the same volumetric properties and aggregate gradation. This is an important issue, but it was beyond the scope of the study.
  
  
• If the compaction temperature must be based on a mastic property and cannot be based on an asphalt binder property, then a procedure will be needed showing how to fabricate a mastic for testing prior to hot-mix plant production. Hot-mix plant production often changes the gradation and surface area of the minus 75-mm material.

I. Results for the Mixtures With Diabase Aggregate and No Hydrated Lime

1. PG 58-28 Unmodified Asphalt Binder

Table 10 shows the data for the PG 58-28 unmodified asphalt binder. The mixing temperature was fixed at 145°C. When the compaction temperature was 157°C, a small amount of smoke was produced during STOA. Therefore, additional specimens were fabricated using a mixing temperature of 165°C and a compaction temperature of 157°C. The mixing temperature should be approximately 165°C when the compaction temperature is 157°C. A large amount of smoke was produced during mixing; therefore, these temperatures cannot be used. Recommended compaction temperatures are from 127 to 147°C.

Table 11 shows that the air voids of the two specimens compacted at 117°C had a difference of 0.9 percent (4.1 vs. 5.0 percent). This is a large difference based on the air voids at the other compaction temperatures. A reason for this was not evident.

2. Novophalt (PG 76-22) Polymer-Modified Asphalt Binder

Table 12 shows the data for Novophalt (PG 76-22). The mixing temperature was fixed at 166°C. When the compaction temperature was 179°C, a large amount of smoke was produced during STOA. Therefore, this compaction temperature cannot be used. Recommended compaction temperatures are from 119 to 159°C, although the data did not provide a firm minimum temperature. Table 12 shows that all of the average air voids were greater than 4.0 percent and the allowable air-void range of 3.5 to 4.5 percent could not be applied to the data. This means that additional asphalt binder would be added during mixture design to obtain a 4.0-percent air-void level. Perhaps the type or grade of asphalt binder would have less effect on the ultimate density if an N-design of 100 or 125 were to be used.

Table 11 shows that the air voids using mixing and compaction temperatures of 166 and 139°C were 1.0 percent apart (5.6 vs. 4.6 percent). These tests were repeated using new materials. Tables 11 and 12 show that the average air-void level for the repeat tests was 4.3 percent, compared to 5.0 percent for the original test. A reason for this was not evident.

Table 12 also shows that the average mixing and compaction temperatures of 180 and 174°C for the mastic could not be used because these temperatures led to a large amount of smoke. There was no smoke or evidence of polymer degradation when preparing the mastics and testing them for viscosity.

3. Styrelf (PG 82-22) Polymer-Modified Asphalt Binder

Table 13 shows the data for Styrelf (PG 82-22). The mixing temperature was fixed at 163°C. When the compaction temperature was 177°C, a small amount of smoke was produced during STOA. Therefore, additional specimens were fabricated using a mixing temperature of 184°C and a compaction temperature of 177°C. The mixing temperature should be approximately 184°C when the compaction temperature is 177°C. A large amount of smoke was produced during mixing; therefore, these temperatures cannot be used. Recommended compaction temperatures are from 147 to 167°C. Table 13 shows that all of the average air-void levels were greater than 4.0 percent. Even so, the allowable air-void range of 3.5 to 4.5 percent was applied to the data.

The average mixing and compaction temperatures for the mastic were 175 and 166°C. The average air-void level of 5.1 percent shown at the bottom of table 13 was significantly greater than the average air-void levels of 4.3 and 4.2 percent using compaction temperatures of 157 and 177°C. Therefore, the compaction tests using 175 and 166°C were repeated using new samples. The new tests provided an average air-void level of 4.3 percent. A reason for the change in the average air-void level was not evident.

Table 11 shows that the air voids for the two specimens using mixing and compaction temperatures of 184 and 177°C, respectively, were 1.2 percent apart (3.8 vs. 5.0 percent). This is a large difference based on the air voids for the other compaction temperatures.

4. Effective Asphalt Binder Contents

Tables 10, 12, and 13 show that the effective asphalt binder contents for the mixtures did not vary significantly with compaction temperature or with the type or grade of asphalt binder.

5. Recompacted Specimens

Table 11 gives the air voids for the mixtures where the compacted specimens were heated, broken apart, and recompacted. The air voids always decreased upon recompaction. Table 14 shows that the compaction process degraded the aggregate slightly. The coarse aggregate used in this mixture had a low L.A. Abrasion of 14. These data indicate that if the air-void level for a specimen is relatively high, the specimen should not be recompacted. New materials must be mixed and compacted.

Previous | Table of Contents | Next